The Internet Company
The Internet Company
Software · Since the internet
All case studies
Insurance·April 2025

Meridian Mutual stopped $6.2M in fraudulent claims and cut claims cycle time 39% with AI fraud detection

A mid-size mutual insurer was paying out an estimated 8% of claims to fraud and couldn't tell which 8%. An AI claims platform that scores every claim for fraud risk, routes suspicious claims to a specialist queue, and auto-settles the clear ones cut leakage, sped up honest claimants, and paid for itself in the first quarter.

InsurTechClaimsFraud DetectionAISIU
Meridian Mutual — Meridian Mutual stopped $6.2M in fraudulent claims and cut claims cycle time 39% with AI fraud detection

This is an illustrative case study. Company names, identifying details, and testimonials have been changed to protect client confidentiality. The technical solution and target outcomes are representative of the work we do. We extend the same confidentiality to every client.

/ The challenge

Where they started

Meridian Mutual handled 84,000 claims a year across auto, home, and commercial lines, and the SIU (special investigations unit) was 4 people reviewing claims flagged by crude rule-based alerts that surfaced 30% of all claims — far more than the team could investigate. The result: an estimated $6.5M a year in fraudulent payouts slipping through, while honest claimants waited 21 days on average for a settlement and 1 in 4 took to social media or the state insurance commissioner. The claims director knew the leakage was real but had no way to find it without hiring a SIU three times the size.

/ The solution

What we built

We deployed our Insure Flow claims platform with an AI fraud-scoring engine trained on 5 years of Meridian's claims history, third-party data enrichment (claims history, provider networks, weather verification), and a triage workflow that auto-settles low-risk claims under a threshold, routes medium-risk claims to adjusters with the AI's reasoning attached, and flags high-risk claims for the SIU with a cited evidence packet. The SIU now works a daily queue of 40 scored claims instead of 25,000 unranked ones. The platform went live in 12 weeks.

/ Results

Target outcomes

$6.2M/yr
Fraudulent payouts stopped
-39%
Claims cycle time
+42%
Fraud detection rate
21 → 9 days
Honest claims settled
30x sharper
SIU caseload quality
/ Gallery

More from this project

We were paying out fraud we knew was there and couldn't find, while punishing honest policyholders with slow settlements. The AI didn't replace our SIU — it gave them 40 claims a day they could actually act on instead of 25,000 they couldn't. The first quarter paid for the whole platform.
Robert Asante
Claims Director, Meridian Mutual

Representative testimonial — name and role changed to protect client identity.